Vasculogenic and Osteogenesis-Enhancing Potential of Human Umbilical Cord Blood Endothelial Colony-Forming Cells§

Authors

  • Yuchun Liu,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    2. BIOMAT, Department of Mechanical Engineering, Faculty of Engineering, National University of Singapore, Singapore
    Search for more papers by this author
  • Swee-Hin Teoh,

    1. Division of Bioengineering, School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Mark S. K. Chong,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Eddy S. M. Lee,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    2. Richard M. Lucas Center for Imaging, Department of Radiology, Stanford University, California, USA
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Citra N. Z. Mattar,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Nau'shil Kaur Randhawa,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Zhi-Yong Zhang,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    2. BIOMAT, Department of Mechanical Engineering, Faculty of Engineering, National University of Singapore, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Reinhold J. Medina,

    1. Centre for Vision and Vascular Science, School of Medicine, Dentistry and Biomedical Science, Queen's University Belfast, Royal Victoria Hospital, Belfast, United Kingdom
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Roger D. Kamm,

    1. Department of Biological and Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, United States of America and Singapore-MIT Alliance for Research & Technology, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Nicholas M. Fisk,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    2. UQ Centre for Clinical Research, University of Queensland, Herston Campus, Brisbane, Australia
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Mahesh Choolani,

    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

  • Jerry K. Y. Chan

    Corresponding author
    1. Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of MedicineNational University of Singapore, Singapore
    2. Department of Reproductive Medicine, KK Women's and Children's Hospital, Singapore
    3. Cancer and Stem Cell Biology Program, Duke-NUS Graduate Medical School, Singapore
    • Experimental Fetal Medicine Group, Department of Obstetrics and Gynaecology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 119228, Singapore
    Search for more papers by this author
    • Author contributions: Y.C.L.: design of the study, collection, analysis and interpretation of data, and manuscript writing; S.H.T. and Z.Y.Z.: conception and design of the study and analysis and interpretation of data; M.S.K.C.: conception and design of the study, analysis and interpretation of data, and manuscript writing; E.S.M.L. and R.D.K.: analysis and interpretation of data; C.N.Z.M.: analysis and interpretation of data and manuscript writing; N.K.R.: provision of study material; R.J.M.B.: provision of study material and analysis and interpretation of data; N.M.F.: analysis of data and manuscript writing; M.C.: conception and design of the study, administrative support, and analysis of data; J.K.Y.C.: conception and design, financial support, administrative support, provision of study materials, analysis of data, manuscript writing, and final approval of the manuscript.

    • Telephone: 65-6772-2672; Fax: 65-6779-4753


  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS June 3, 2012.

Abstract

Umbilical cord blood-derived endothelial colony-forming cells (UCB-ECFC) show utility in neovascularization, but their contribution to osteogenesis has not been defined. Cocultures of UCB-ECFC with human fetal-mesenchymal stem cells (hfMSC) resulted in earlier induction of alkaline phosphatase (ALP) (Day 7 vs. 10) and increased mineralization (1.9×; p < .001) compared to hfMSC monocultures. This effect was mediated through soluble factors in ECFC-conditioned media, leading to 1.8–2.2× higher ALP levels and a 1.4–1.5× increase in calcium deposition (p < .01) in a dose-dependent manner. Transcriptomic and protein array studies demonstrated high basal levels of osteogenic (BMPs and TGF-βs) and angiogenic (VEGF and angiopoietins) regulators. Comparison of defined UCB and adult peripheral blood ECFC showed higher osteogenic and angiogenic gene expression in UCB-ECFC. Subcutaneous implantation of UCB-ECFC with hfMSC in immunodeficient mice resulted in the formation of chimeric human vessels, with a 2.2-fold increase in host neovascularization compared to hfMSC-only implants (p = .001). We conclude that this study shows that UCB-ECFC have potential in therapeutic angiogenesis and osteogenic applications in conjunction with MSC. We speculate that UCB-ECFC play an important role in skeletal and vascular development during perinatal development but less so in later life when expression of key osteogenesis and angiogenesis genes in ECFC is lower. Stem Cells2012;30:1911–1924

INTRODUCTION

Circulating endothelial progenitor cells (EPCs) isolated from human peripheral blood (PB) have been implicated in the process of neovascularization [1] and investigated in various vascular paradigms. Numerous clinical trials using EPCs have been performed, largely for treatment of limb ischemia [2] and myocardial infarction [3]. However, unlike preclinical rodent models, only modest benefits have been observed; for example, left ventricular ejection fraction improved only 2%–8%, with no reduction in death or stroke [3]. A major problem confronting this field is the inconsistency in defining what constitutes an EPC, with the use of CD34+, CD133+, and VEGFR2+ among others as markers of EPC, leading to high levels of contamination with hemopoietic stem and progenitor cell types, and few studies providing evidence of functional differentiation in vitro or in vivo [4]. Furthermore, few protocols exist that can reliably expand these putative EPCs into clinically useful numbers.

Recently, the use of a stringent culture system for deriving PB or umbilical cord blood (UCB) EPC termed endothelial colony-forming cells (ECFC) allowed the reliable isolation and expansion of cells that are clonogenic, have a well-defined immunophenotype, and the ability to engraft functionally as blood vessels in murine models [4, 5]. These ECFC are relatively pure and uncontaminated by hemopoietic lineages, being CD45 negative, and can be reliably isolated and expanded, raising the possibility of their use in regenerative medicine.

In previous work, we have shown the superiority of human fetal-mesenchymal stem cells (hfMSC) over other MSC sources for generation of highly osteogenic tissue-engineered bone grafts using bioresorbable osteoinductive macroporous scaffolds [6]. Here, we propose the use of UCB-derived-ECFC to enhance the vasculogenic potential of hfMSC in coculture. In this study, we examined the interactions of well-characterized UCB-ECFC with hfMSC through a series of in vitro and in vivo transplantation assays, with the overall goal of defining the use of ECFC in the generation of vascularized grafts, of relevance not only to bone but also to other fields of tissue engineering where large-voluminous grafts are required.

MATERIALS AND METHODS

Samples, Animals, and Ethics

Human tissue collection for research purposes was approved by the Domain Specific Review Board of National University Hospital Systems (DSRB-D-06-154), in compliance with International Guidelines regarding the use of fetal tissue for research as previously described [6]. In all cases, patients gave separate written consent for the use of the collected tissue. Fetal femurs were collected for isolation of hfMSC after clinically indicated termination of pregnancy. In this study, a sample derived from a 17 weeks gestation was used. Human UCB samples from newborns were collected from normal term deliveries (n = 8). mRNA derived from UCB-ECFC (n = 1) and adult PB-ECFC samples (n = 2) were obtained from Belfast, U.K.

Male nonobese diabetic severe combined immunodeficient (NOD SCID) mice (6–8 weeks old) were acquired from Animal Resources Centre Canning Vale, Western Australia, www.arc.wa.gov.au/contact.php (Australia). All procedures were approved by the Institutional Animal Care and Use Committee at The National University of Singapore, and procedures were performed in the association for assessment and accreditation of laboratory animal care international (AAALAC) accredited animal facilities.

Isolation, Culture, and Characterization of hfMSC and ECFC

hfMSC were isolated from bone marrow as previously described [7]. Briefly, single-cell suspensions were prepared by flushing the bone marrow cells from femurs using a 22-gauge needle, passed through a 70 μm cell strainer (BD Biosciences, San Jose, California, http://www.bdbiosciences.com/home.jsp), and plated on 10-cm diameter plates (NUNC, Thermo Scientific, Rochester, NY, www.nuncbrand.com/us/page.aspx?ID=1238) at 106 cells per milliliter. Adherent spindle-shaped cells were recovered from the primary culture after 4–7 days. Nonadherent cells were removed with initial media changes every 2–3 days. At subconfluence, they were trypsinized and replated at 104 cells per square centimeter. MSCs were cultured in Dulbecco's modified Eagle's medium-Glutamax supplemented with 10% fetal bovine serum (FBS), 50 U/ml penicillin, and streptomycin (GIBCO Life Technologies, Grand Island, New York, http://www.invitrogen.com/site/us/en/home/support/Contact-Us.html), thereafter referred to as D10 media. Cells from passages 2–4 were characterized as previously described [6].

UCB-derived ECFC were obtained as previously described [5]. Briefly, blood was diluted and overlaid onto Ficoll-Paque PLUS (Amersham, Piscataway, New Jersey). Mononuclear cells from the buffy coat retrieved from UCB by low-density centrifugation were resuspended in endothelial growth media (EGM) (Cambrex Charles City, Iowa, http://www.cambrex.com/locations/view/charles_city) supplemented with 10% FBS, thereafter referred to as EGM10, and then plated onto 10-cm diameter tissue culture dishes coated with type I rat tail collagen (BD Biosciences, Bedford, MA) at 37°C, 5% CO2 in a humidified incubator. Culture media were changed every 3 days subsequently. Typical cobblestone colonies appeared after 2 weeks and were subcultured at subconfluence.

Other Media Preparation

Well-defined bone media (BM) were prepared using D10 media supplemented with bone inducing elements—10 mM β-glycerophosphate, 10−8 M Dexamethasone, and 0.2 mM Ascorbic acid (Sigma-Aldrich, St Louis, Missouri, http://www.sigmaaldrich.com/united-states.html/).

To obtain ECFC-conditioned media (ECFC-CM), ECFC were cultured in EGM10 until confluence and washed thrice with phosphate buffered saline (PBS) prior to BM addition. Media were collected after 72 hours, centrifuged, filtered through a 0.22 μm filter, and further diluted in ratios of 1:1 and 1:10 with BM, thereafter referred to as ECFC-CM (1:1) and ECFC-CM (1:10), respectively.

Osteogenic Assays

Calcium deposition was assayed as previously described [6]. Briefly, each sample well (n = 3) was incubated with 0.5 N acetic acid overnight to dissolve calcium. A colorimetric calcium assay kit (BioAssay Systems, Hayward, California, http://www.bioassaysys.com/contact_us.php) was used to quantitate calcium content spectrophotometrically at 612 nm according to manufacturer's instruction.

Alkaline phosphatase (ALP) was assayed as previously described [6]. Briefly, samples (n = 3) were gently rinsed twice with PBS, incubated in 1 mg/ml of Collagenase type I (Sigma-Aldrich) dissolved in 0.1% Trypsin (Life Technologies, Grand Island, New York, http://www.invitrogen.com/site/us/en/home/Products-and-Services/Applications/Bioproduction/Why-Us.html) solution, and incubated at 37°C for 2 hours to digest the extracellular matrix completely. After three cycles of freeze-thaw, the cell lysate solution was assayed for ALP activity using SensoLyte pNPP Alkaline Phosphatase Assay Kit (AnaSpec, Fremont, California, http://www.bionity.com/en/companies/17292/anaspec-inc.html) following the manufacturer's instruction.

Von Kossa staining was performed as previously described [6]. Briefly, samples were gently rinsed twice with PBS then fixed with 10% formalin for 1 hour, followed by two washes with distilled water and stained with freshly made 2% silver nitrate (Sigma-Aldrich, St Louis, Missouri) in distilled water (wt/vol) for 10 minutes in the dark and expose to sunlight for 30 minutes.

Gene Expression from Microarrays

Two microarray analyses were performed. In the first, total RNA was extracted from UCB-ECFC grown to subconfluence in biological triplicates, using the RNeasy kit (Qiagen, Valencia, CA) as per the manufacturer's protocol. Ten micrograms total RNA was used to generate labeled cRNA and hybridized to Human Genome U133 Plus 2.0 arrays (Affymetrix, Santa Clara, CA). Data files were imported into GeneSpring GX 11 (Agilent Technologies, Palo Alto, California). In the second microarray analysis, data files of UCB-ECFC (gse508535) and adult PB-derived ECFC (PB-ECFC) (gsm508539-41) samples from Gene Expression Omnibus Series gse20283 (www.ncbi.nlm.nih.gov/geo) as published by Medina et al. [8] were imported into GeneSpring GX 11. Data files from both microarray studies were subjected to normalization, summarization, and gene ontology (GO) analysis with GeneSpring GX 11. Probe sets related to osteogenesis were identified by the related GO terms (GP:0001503 and children) [9]. In the analysis of ECFC transcriptome, selected probe sets were sorted in descending order of the mean expression of triplicate samples and color-coded with the highest red, median (4.59) yellow, and lowest blue. Further details are described in Supporting Information data.

Quantitative PCR Analysis

Extracted total RNA of ECFC derived from UCB and human adult PB were supplied by Belfast, U.K. Briefly, 1 μg of RNA was converted to cDNA reverse-transcribed using the Superscript II First Strand cDNA synthesis kit (Life Technologies, Austin, Texas, http://www.invitrogen.com/site/us/en/home/Products-and-Services/Applications/Bioproduction/Why-Us.html) in a total volume of 19 μl. Quantitative PCR reactions were performed in triplicates for the following osteogenic genes: BMP1, BMP4, BMP6, COL1A1, TGF-β1, and Osteonectin. Two microliters of cDNA, 10 μl of SYBR Green PCR Master Mix (Applied Biosystems, Carlsbad, California, http://www.appliedbiosystems.com/absite/us/en/home/contact-us.html), and 1 μl of 5 μM primers formed the working solution that was topped up to a final volume of 20 μl. Thermal cycle conditions were 95°C for 10 minutes, then 45 cycles at 95°C for 15 seconds, and 60°C for 1 minute. Amplifications were monitored with the ABI Prism 7000 (Applied Biosystems, Carlsbad, California, http://www.appliedbiosystems.com/absite/us/en/home/contact-us.html). Results were normalized against the housekeeping gene β-actin, and relative gene expression was analyzed with the 2-ddCt method. The primers used are as follows:

Table  . 
original image

Antibody Array and Microarray Scanning

Semiquantitative G series antibody arrays 6, 7, and 8 (Human Cytokine Antibody Array [RayBiotech, Norcross, Georgia, http://www.raybiotech.com/contact_us.asp]) were used to detect 174 cytokines as per manufacturer's instructions. Briefly, after blocking the arrays with blocking buffer for 30 minutes, they were incubated with either ECFC-CM or BM for 2 hours at room temperature. Samples were decanted from each well and washed three times with wash buffer, incubated with biotin-conjugated antibodies, and followed by fluorescent dye-conjugated streptavidin for 2 hours each. Thereafter, arrays were washed, centrifuged, and allowed to dry before scanning at an excitation frequency of 532 nm (Axon Genepix 4000B). Background signals were obtained from the negative control, bovine serum albumin (BSA) were subsequently subtracted from median values, and sample averages were normalized against the positive control, biotinylated protein as supplied by the manufacturer. Signals from ECFC-CM were then subtracted from BM values and analyzed.

Preparation of Cellular-Scaffold Constructs

Three-dimensional (3D) polycaprolactone-tricalcium phosphate (PCL-TCP) (80/20) scaffolds (Osteopore International, Singapore) with a lay-down pattern of 0/60/120°, porosity of 70%, and average pore size of 0.523 mm were used in 4 × 4 × 4 mm3 dimensions. These were treated with 5 M NaOH for 3 hours to enhance their hydrophilicity and washed thoroughly with PBS thrice prior to ethanol sterilization.

For cell loading, ECFC/hfMSC (1:1) and hfMSC were suspended in Tisseel Fibrin Sealant (Baxter, Zurich, Switzerland) before seeding into the porous scaffolds, keeping the hfMSC density constant (at 3,000 cells per cubic millimeter) and maintained in BM culture in 24-well plates for 4 days prior to implantation. In some experiments, differentially labeled cells (green fluorescent protein [GFP]-ECFC and H2B-mCherry-hfMSC) were prepared as described below before loading onto scaffolds and maintained in BM for 14 days.

Lentiviral-Transduction of EPC and hfMSC

The lentiviral vectors were produced as described previously [10]. Briefly, the transfer plasmids (PGK-H2BmCherry and pRRLSIN.cPPT.PGK-GFP.WPRE) were cotransfected with pMD2.G and pCMV.ΔR8.74 into HEK293T cells. The supernatant was collected at 48 and 72 hours following transfection and concentrated by two rounds of ultracentrifugation at 50,000g for 2 hours and the final pellet was dissolved in a small volume of 1% BSA in PBS (1/100 of starting volume). The number of transducing units (TUs) of the vectors was determined by infecting 100,000 293T cells using serial dilution of the vector. The dilution resulting in <30% GFP or red fluorescent protein (RFP) RFP-positive cells was used to calculate the number of TUs per milliliter. For transduction, cells were seeded at 0.5 × 104 cells per square centimeter in T-25 flasks and exposed to lentivirus with 4 mg/ml polybrene at multiplicity of infection of five. GFP-labeled EPC and nuclear-mCherry-labeled hfMSC at >90% transduction efficiencies were used in the experiments.

Imaging

Cellular morphology and adhesion were examined daily by phase-contrast light microscopy and imaged at fixed time intervals.

Monolayer and 3D cultures were viewed and scanned under a confocal laser microscope using laser wavelengths of 405 nm (green), 594 nm (red), and 4′,6-diamidino-2-phenylindole (DAPI) channel (blue) (Olympus, FV300 Fluoview, Japan). Image analysis software Imaris and ImageJ were used for the quantification of lamin A/C staining and vessels in at least three different regions of the cryosections, respectively.

In Vivo Transplantation and Assays

Under general anesthesia, cellular-scaffold constructs were implanted into subcutaneous pockets generated at the dorsal surface of each immunodeficient mouse and the skin closed with 5-O Vicryl sutures. Scaffolds were harvested at 3 and 8 weeks for histological analysis.

Vascularization Assay

Visual observation of vascularization was performed with the use of a vascular contrast, Microfil (MV-120, Flow Tech, Carver, Massachusetts). Briefly, mice were anesthetized, heparinized with 0.1 ml of 500 U via the tail vein, and their left ventricles were cannulated before perfusion with 50 ml of 10 U heparinized saline. Upon exsanguination, 10 ml of Microfil solution along with MV-curing agent (10% of total volume) was perfused. Animals were then stored at 4°C for 2 hours for the silicon compound to polymerize before scaffolds were harvested and fixed in 4% paraformaldehyde as previously described [6].

Histology

Cellular-scaffold constructs were placed on crushed dry ice, embedded in optimal cutting temperature compound, and sectioned at 50-μm thickness with a Cryostat (CM 3050S, Leica, Wetzlar, Germany).

Immunohistochemistry

Cryosections from each sample were blocked with 5% normal goat serum for 1 hour and left to react with monoclonal mouse anti-human lamin A/C (1:100, Vector Laboratories, Peterborough, United Kingdom, http://www.vectorlabs.com/contactus.aspx), mouse anti-human CD31 (1:100, Millipore, Billerica, Massachusetts, http://www.millipore.com/company/flx4/about), and rat anti-mouse CD31 (1:100, BD Pharmingen, San Jose, California) antibodies, respectively, overnight; sections were then incubated with goat anti-mouse secondary antibodies (1:100 Alexafluor 488/594 Life Technologies, Auckland, New Zealand, http://www.invitrogen.com/site/us/en/home/Products-and-Services/Applications/Bioproduction/Why-Us.html for 90 minutes and counterstained with propidium iodide or DAPI mounting media. Images were visualized through confocal microscopy as above.

Statistical Analysis

Parametric data are presented as mean ± standard deviation and were analyzed by t test or analysis of variance (ANOVA). *, p < .05; **, p < .01; ***, p < .001 indicate significant differences between samples.

RESULTS

Derivation of UCB-ECFC and hfMSC

UCB-ECFC emerged as polygonal cells in a cobblestone pattern which expressed CD31, CD144, and vWF, took up acetylated-low density lipoprotein (LDL), and differentiated into tube-like networks when plated on Matrigel (Supporting Information Fig. S1) [11]. hfMSC isolates grew as spindle-shaped cells which were negative for hemopoietic and endothelial markers (CD14, CD34, and CD45) and were positive for CD73, CD105, and CD90 using flow cytometry analysis (Supporting Information Fig. S2). They had a colony-forming unit-fibroblast (CFU-F) efficiency of 75.1% ± 5.0% and differentiated into osteoblasts, adipocytes, and chondrocytes in permissive media [7, 12].

ECFC Monoculture Does Not Undergo Osteogenic Differentiation in BM

To determine suitable culture conditions for stimulating osteogenic differentiation, we cultured EPC and hfMSC separately in BM, EGM10, and EGM10 supplemented with bone-inducing elements (b-EGM10). Only BM and not EGM10 or b-EGM10 induced osteogenic differentiation of hfMSC, seen as extracellular crystals under phase-contrast microscopy (Fig. 1A) and confirmed by Von Kossa staining (Fig. 1B). Quantitatively, BM induced high levels of calcium deposition in hfMSC (24.1 μg per well), but not in ECFC (0.1 μg per well; p < .001) cultured in BM, while no calcium was deposited in either cell type when cultured in EGM10 or b-EGM10 (Fig. 1C).

Figure 1.

BM is necessary to induce osteogenic differentiation of hfMSC. (A): hfMSC laid down extracellular minerals when cultured in BM while cell debris was observed with ECFC in BM. Cells cultured in EGM10 and b-EGM10 did not demonstrate mineralization. (B): hfMSC cultured in BM showed dark Von Kossa stains for calcium crystals but none in the other study groups. (C): This was confirmed by quantitative calcium assays with only hfMSC grown in BM laying down significant amounts of calcium. All assays were performed on Day 14. Abbreviations: BM, bone media; ECFC, endothelial colony-forming cells; hfMSC, human fetal-mesenchymal stem cells; ***, p<0.001.

ECFC/hfMSC Coculture Enhanced Osteogenic Induction in hfMSC

To determine the optimal cell-ratio for osteogenic induction, ECFC and hfMSC were cocultured at different ratios but a constant hfMSC seeding density (20,000 cells per square centimeter). Cocultured ECFC/hfMSC resulted in earlier mineralization compared to hfMSC alone, with extracellular crystals seen by Day 4 in coculture groups compared with Day 7 in hfMSC-only cultures (Fig. 2A). ECFC/hfMSC cocultured in a 1:1 ratio achieved the earliest ALP peak on Day 7 compared to other groups (Fig. 2B) and a 1.9-fold higher calcium deposition on Day 14 compared to hfMSC cultures (p = .001, n = 3 replicates) (Fig. 2C), indicating enhanced osteogenic differentiation.

Figure 2.

ECFC/hfMSC in coculture induces earlier osteogenic differentiation of hfMSC. (A): ECFC/hfMSC groups in varying ratios displayed earlier mineralization compared to ECFC and hfMSC monocultures. (B): Quantitative ALP measurements with ECFC/hfMSC (1:1) demonstrated the earliest ALP peak activity on Day 7. (C): ECFC/hfMSC (1:1) cocultures laid down the most calcium on Day 14. ***, p < .001. Abbreviations: ALP, alkaline phosphatase; ECFC, endothelial colony-forming cells; hfMSC, human fetal-mesenchymal stem cells.

ECFC Potentiates Osteogenesis of hfMSC Through Secreted Factors

To investigate the mechanism for the observed osteogenic enhancement of hfMSC, we added ECFC-CM directly into hfMSC cultures. This resulted in more extracellular crystals being deposited than in hfMSC cultured in BM alone, as seen on light microscopy (Fig. 3A). This was confirmed by increased Von Kossa staining on Day 14 in ECFC-CM (1:1) and ECFC-CM (1:10) cultures compared to BM cultures (Fig. 3B). Quantitatively, ALP activity peaked earlier in the EPCCM (1:1) group at Day 7 compared to Day 10 for ECFC-CM (1:10) and BM groups, with peak values being 2.2 and 1.8-fold higher when cultured in ECFC-CM (1:1) and ECFC-CM (1:10), respectively, compared to BM (Fig. 3C). Correspondingly, 1.3-fold more calcium was deposited in the ECFC-CM (1:1) compared to BM by Day 14 (p < .05) (Fig. 3D). Collectively, these results suggest that soluble factors secreted by ECFC play a role in potentiating osteogenic differentiation of hfMSC in a dose-dependent manner.

Figure 3.

Dose-dependent effect of soluble factors in ECFC-CM enhanced osteogenic differentiation of hfMSC. (A): hfMSC cultured in ECFC-CM (1:1) and ECFC-CM (1:10) demonstrated earlier and more extensive mineralization than when cultured in BM. (B): Greater Von Kossa staining was observed in ECFC-CM (1:1) and ECFC-CM (1:10) compared to BM at Day 14 of osteogenic differentiation. (C): ECFC-CM (1:1) and ECFC-CM (1:10) displayed higher (2.2-fold and 1.8-fold, respectively) peak of ALP activity than BM. (D): Calcium deposited in ECFC-CM (1:1) and ECFC-CM (1:10) was consistently higher (1.5-fold and 1.4-fold, respectively) than BM in all time points. (E): ALP levels were consistently higher in BM than in basal media, D10, with the coculture displaying highest levels on Day 14. (F): Both hfMSC and ECFC/hfMSC cocultures required osteogenic supplements to induce osteogenic differentiation, which was potentiated with coculture, with deposition of calcium after 14 days of induction, further confirmed by Von Kossa staining *, p < 0.05 ***, p < 0.01. (G). ***, p < .001. Abbreviations: ALP, alkaline phosphatase; BM, bone media; CM, conditioned media; ECFC, endothelial colony-forming cells; hfMSC, human fetal-mesenchymal stem cells.

Next, we assessed the ability of ECFC themselves in the absence of bone-inducing components to induce osteogenic programming of hfMSC upon direct contact. hfMSC and ECFC/hfMSC groups cultured in BM showed twofold (p < .01) and 2.5-fold (p < .001) higher ALP activity, respectively, compared to culture in basal D10 media (Fig. 3E). Similarly, calcium deposition was found only in BM cultures but not in D10 cultures for either hfMSC and ECFC/hfMSC, as assayed by Von Kossa staining and quantitated (Fig. 3F, 3G), suggesting that the addition of ECFC to hfMSC in basal D10 media does not induce osteogenic programming in hfMSC, but rather, ECFC potentiates osteogenic differentiation of hfMSC cultured in BM.

UCB-ECFC Secrete a Broad Spectrum of Proteins, Including Bone-Related Proteins from the TGF-β Superfamily

Two approaches were adopted to interrogate the identity of secreted proteins in ECFC-CM responsible for the enhanced osteogenic differentiation of hfMSC. First, we used a transcriptomic approach to find that ECFC highly expressed a range of genes associated with angiogenesis as well as genes associated with osteogenesis such as members of the TGF-beta pathway including SMAD1, 2, 3, and 5 and bone morphogenetic proteins, BMP-1, 2, 3, 6, 7, and 8 (Supporting Information Table S1A, S1B, S2, S3).

Next, we used a semiquantitative antibody array (Supporting Information Table S4) to assay the secretome of ECFC found in ECFC-CM. ECFC secreted several inflammatory cytokines, angiogenic factors, as well as bone-related proteins, all which play an important role in regulating the biological processes involved in fracture healing. High levels of secreted proteins such as Angiogenin, Ang-2, GRO, IL-6, IL-8, MCP-1, MIP-3a, platelet derived growth factor-BB (PDGF-BB), and TIMP-2 were identified (Supporting Information Table S5A). In addition, we confirmed our transcriptomic observations identifying bone-related proteins, BMP-5, BMP-6, BMP-7, TGF-β1, and TGF-β2 from the TGF-β superfamily secreted into ECFC-CM. These, along with basic fibroblast growth factor (bFGF), Endoglin, FGF-4, IGF-1, MCP-1, Oncostatin, Osteoprotegerin, PDGF, and vascular endothelial growth factor (VEGF) found in the ECFC-CM, may have contributed directly or indirectly to the observed enhancement in osteogenesis (Supporting Information Table S5B).

ECFC Derived from UCB Are More Osteogenic and Angiogenic Compared to Adult PB-Derived ECFC

We validated this by reanalyzing transcriptomic data from a previously published study of the effect of EPCs and outgrowth endothelial cells (ECs) on osteogenesis and angiogenesis processes that reported data separately for well-characterized UCB-ECFC and adult PB-ECFC (Fig. 4A) [8]. This showed that UCB-ECFC expressed BMP-1 (4.1×), BMP-2 (2.9×), BMP-4 (2.4×), BMP-6 (1.5×), TGF-β2 (11.0×), and COL1A1 (2.4×) among others at higher levels than in adult PB-ECFC. Using different samples of UCB (n = 1) and adult PB-ECFC (n = 2), we validated some of these genes through quantitative PCR to show that BMP1, BMP4, BMP6, COL1A1, TGF-β1, and osteonectin were upregulated in UCB over adult PB-ECFC (Fig. 4B). Similarly, UCB-ECFC expressed key angiogenesis genes at higher levels (Fig. 4A and Supporting Information Table S6, S7).

Figure 4.

UCB versus adult PB-derived ECFC taken from Medina et al. [8] demonstrated higher expression levels of key osteogenic and angiogenic genes using (A) transcriptomic microarray analysis and (B) quantitative real-time PCR. Abbreviations: ECFC, endothelial colony-forming cell; PB, peripheral blood; UCB, umbilical cord blood.

ECFC Within hfMSC Cultures Form Islets and Tubular Structures

By labeling ECFC with GFP, we observed poor viability of ECFC when cultured in D10 over 14 days, whereas the addition of the standard osteogenic-inducing agents dexamethasone, ascorbate, and β-glycerophosphate improved their viability markedly (Fig. 5A). Coculture of GFP-labeled ECFC with nuclear RFP-labeled hfMSC resulted in the formation of ECFC-derived islets that were evolved by Day 7 into tube-like structures when cultured in BM but not when cultured in D10 (Fig. 5B).

Figure 5.

ECFC potentiate osteogenic programming of hfMSC and in vitro tubule formation in the presence of bone-inducing components. (A): Green fluorescent protein (GFP)-labeled ECFC showed better survival when cultured in BM than in D10. (B): Coculture of GFP-ECFC and H2B-RFP-hfMSC (red fluorescence nuclear staining) resulted in the formation of ECFC islets (white arrowheads), leading eventually to the development of tubular structures (yellow arrows) by Day 7. (C): Formation of GFP-ECFC vessel-like structures with complex branching points was observed when ECFC/hfMSC were cocultured for more than 14 days within a 3D culture system within a macroporous scaffold. Abbreviations: BM, bone media; ECFC, endothelial colony-forming cells; hfMSC, human fetal-mesenchymal stem cells.

Next, we looked at the ability of this coculture system to induce tube-like structures in a 3D culture system within a macroporous scaffold. Cocultured GFP-ECFC and hfMSC embedded in fibrin and loaded onto these scaffolds proliferated and occupied the porous scaffold over time, with complex tubular structures with multiple branch-points seen throughout the scaffold by Day 14 (Fig. 5C).

ECFC/hfMSC Cocultures Induced More Robust Neovasculogenesis and Ectopic Bone Formation In Vivo

Following the robust formation of vascular structures within 3D ECFC/hfMSC cocultures in vitro, we implanted ECFC/hfMSC and hfMSC-loaded scaffolds subcutaneously into immunodeficient mice. Three weeks after implantation, perfusion of a vascular contrast agent showed an extensive network of vessels surrounding the harvested ECFC/hfMSC scaffolds, while comparatively few vessels were seen on hfMSC-only scaffolds (Fig. 6A).

Figure 6.

In vivo neovasculogenesis and ectopic bone formation. (A): Increased vascularization of the ECFC/hfMSC scaffolds was evident 3 weeks after implantation, as seen after Microfil perfusion (blue vessels). (B): At 8 weeks, human blood vessels stained with human-specific CD31 (green) were seen coursing through ECFC/hfMSC scaffolds but not hfMSC scaffolds. These human vessels can be seen enmeshed with murine vessels (stained red with murine CD31 antibody) as evident in a 50-μm section (merged and stacked confocal images (C)). ECFC/hfMSC scaffolds contained a 2.2-fold (p = .001) higher density of host-derived murine-CD31-positive vessels (red) compared to hfMSC scaffolds (arrows indicating area of human-murine vasculature junctures) (D), while human vessels constituted 30.2% of the total vessel area within the construct (E). (F): Immunostaining with human Lamins A/C demonstrated high levels of chimerism in both scaffolds, with a trend toward lower human cell chimerism in ECFC/hfMSC scaffolds compared with hfMSC scaffolds. (G, H): Von Kossa staining of the implants showed darker level of staining (scaffold regions has been denoted as S) and a slightly higher level of calcium deposited in ECFC/hfMSC scaffolds compared to hfMSC scaffold. *, p < .05. Abbreviations: BM, bone media; ECFC, endothelial colony-forming cells; EPC, endothelial progenitor cell; hfMSC, human fetal-mesenchymal stem cells; PI, propidium iodide.

By Week 8, human CD31-positive blood vessels with complex networks could be seen coursing through ECFC/hfMSC scaffolds but not through hfMSC-only scaffolds (Fig. 6B). These human vessels were seen to form chimeric vessels, joining with host-derived vessels stained with murine-specific CD31 antibodies (white arrows in Fig. 6C) at multiple levels throughout the scaffold (Fig. 6B, 6C). By staining for murine-specific CD31-positive blood vessels (red), we found that ECFC/hfMSC scaffolds developed 2.2-fold greater area of murine-specific CD31 per square micrometer than hfMSC-only scaffolds (p = .001), suggesting a higher degree of host-derived neovascularization (Fig. 6D). In ECFC/hfMSC scaffolds, human vessels accounted for 30.2% of all vessels within the scaffold core (Fig. 6E). This observation was supported by a lower degree of human cell chimerism in the ECFC/hfMSC scaffolds compared to hfMSC-only scaffolds (55.9% ± 4.7% vs. 74.8% ± 12.3%, p > .05) (Fig. 6F) and may reflect a higher degree of host-derived cellular infiltrate. Von Kossa staining revealed darker staining in ECFC/hfMSC scaffolds (Fig. 6G), suggesting greater osteogenic differentiation was induced in the cocultured scaffolds. This was verified by the calcium quantification that showed a trend toward greater ectopic bone formation in the coculture group (Fig. 6H), implicating the increased vascularization in enhancing osteogenic differentiation in vivo.

DISCUSSION

Several groups have cocultured putative EPC with MSC or osteoblast-like cell types on the premise that this strategy will generate prevascular networks within bone tissue-engineered grafts to aid bone repair [13–18]. However, the majority of studies used ECs [19–23] rather than ECFC. In this project, well-characterized UCB-ECFC were cocultured with hfMSC [7] chosen primarily for their ability to undergo a well-defined osteogenic differentiation pathway [7, 24, 25] and superiority in osteogenic differentiation over adult counterparts [7, 25] for bone tissue engineering applications. We unexpectedly found a twofold enhancement in the osteogenic differentiation of hfMSC when cocultured with UCB-ECFC, and showed that this effect was at least in part through paracrine signaling involving secreted members of the TGF-β superfamily. In addition, the addition of UCB-ECFC to hfMSC led to the formation of tube-like network of endothelial structures in vitro, with the generation of chimeric human-murine blood vessels and enhancement of host-neovascularization in vivo. Furthermore, we provide evidence that there is an ontological advantage in skeletal morphogenesis and angiogenesis of the more primitive UCB-derived ECFC over adult sources.

EPC were first identified in PB [1] and subsequently isolated from other sources including bone marrow, fetal liver, and UCB [4]. However, the isolation of EPC from UCB is still favored over other sites due to its plentiful supply and noninvasive collection process. In addition, EPC in UCB are found at higher frequencies and have higher proliferative and vasculogenic capacity than adult PB-EPC [26].

Studies of EPC have largely focused on their vasculogenic potential, with evidence of efficacy in augmenting neovascularization in several different ischemic models. Recently, there has been increasing interest in the use of EPC populations for bone repair in mice [27], rat [28], and ovine models [29]. Our key finding was a potentiated osteogenic response of hfMSC in vitro via paracrine signaling, where soluble factors present in ECFC-CM exerted a profound effect on osteogenic differentiation of hfMSC. The probable mechanism mediating the enhanced osteogenic effect is likely to be soluble factors secreted by ECFC when cocultured in BM, given the lack of osteogenic induction when the coculture was performed in basal hfMSC growth media. Bone inductive components in BM could have been responsible for inducing osteo-inductive secretions in ECFC as detected by the protein array blot that identified a milieu of secreted factors present in ECFC-CM, with negligible levels of these proteins found in BM alone (data not shown). The bone-related morphogens detected included BMP-5, BMP-6, BMP-7, TGF-β1, and TGF-β2, which are known key facilitators or inductors of osteogenic programming in MSC. BMPs and TGF-βs of the TGF-β superfamily are produced by osteoblasts and other bone cells incorporated into mineralized matrix, thus contributing to osteoblast differentiation in vitro [30]. In vivo, such growth factors participate actively in various stages of intramembranous and endochondral bone ossification for bone formation and remodeling [31]. In particular, the potent osteo-inductive abilities of BMPs have been widely demonstrated in various animal and clinical studies [32] and more recently, BMP-2 has also been used as an adjunct to the standard of care in the BMP-2 evaluation in surgery for tibial trauma study involving 450 patients with diaphyseal open tibial fractures [33].

Despite the low relative levels of BMP detected, ECFC-CM were still able to induce a significant increase in osteogenic differentiation of MSC in a dose-dependent manner. In addition, VEGF detected in ECFC-CM may have contributed significantly to the observed phenomena. VEGF has been reported to induce ALP activity in osteoblasts [34] and enhance osteoblast differentiation [35]. Behr et al. also demonstrated that VEGF-A promotes osteogenic and endothelial differentiation [36], thus playing a significant role in skeletal repair [37].Other bone-morphogenic factors such as FGFs, PDGF, oncostatin M, and endoglin have also been reported to induce osteoblast differentiation. The effects of multiple osteogenic factors secreted by ECFC could have acted in synergy to enhance the potency of osteogenic differentiation of MSC, leading to the strong enhancement in mineralization observed in vitro.

Few groups have investigated the mechanism behind these enhanced osteogenic effects. Dohle et al., through a study of primary osteoblast and adult PB late endothelial outgrowth cocultures, implicated the sonic hedgehog (Shh) pathway as a key modulator in EPC-enhanced osteogenesis through supplementation of Shh and its inhibitor [38]. Saleh et al. in demonstrating enhanced proliferation and osteogenic differentiation of human MSC exposed to CM from human umbilical vein endothelial cells (HUVEC) alluded to the possibility of FGF, Wnt, BMP, and Notch pathways as possible regulators of this phenomenon [19], albeit without identifying these putative regulators directly. Wang et al. showed that cocultured HUVECs augmented osteogenic differentiation of adipose-derived MSC, with BMP-2 being identified in HUVEC-CM, although they did not test its paracrine activity [39]. In contrast to our suggested mechanism of paracrine signaling, several other groups failed to show osteogenic enhancement when MSCs were cultured in the CM of ECs [40], instead suggesting direct cell-cell contact via gap junction communication was required to mediate the osteogenic enhancement effect [21, 41]. These groups, however, used mature human ECs in coculture with osteoblast-like cells, which although they have shown promise in generating a prevascularized network [15, 23] and/or enhancing osteogenesis [20, 21, 40–43], their biological phenotype depends upon the site of the endothelium from which they were harvested from [44].

The use of BM to stimulate osteogenic differentiation of MSC while maintaining high survivability of EPC without altering phenotypic stability [18] is essential for the observed enhanced osteogenic differentiation. ECFC survived in monoculture at high levels of confluence in BM while maintaining their cobblestone morphology in the absence of any standard endothelial growth factors supplements (VEGF, EGF, bFGF, IGF-1, ascorbic acid, hydrocortisone, and heparin). This is likely due to the presence of dexamethasone in BM, a potent synthetic glucocorticoid critical for EPC survival as an alternative to the hydrocortisone supplemented in EGM. In contrast to our findings, others have suggested that growth factor supplements are necessary to sustain EPC survival. Usami et al. attributed the difficulties of coculturing EPC and MSC in direct contact to the inability of EPCs to survive in osteogenic media and resorted to culturing the cells on polylactide-coated collagen fiber mesh separately in their respective growth media prior to coimplantation in vivo [14]. Other similar studies maintained cocultures in EGM instead [17, 18, 45]. In our study, UCB-ECFC were unable to induce osteogenic differentiation on hfMSC in the absence of bone inductive components of dexamethasone, ascorbate, and β-glycerophosphate.

By comparing the transcriptome of well-defined ECFC populations derived from primitive UCB with mature adult PB-ECFC from a previous study [8], both osteogenic and vasculogenic genes were significantly more expressed in UCB compared to adult PB-ECFC. While the higher vasculogenic potential of UCB-derived ECFC is largely expected, the higher expression of osteogenesis-related genes here may reflect the intense skeletal development occurring during neonatal life as compared to the mature adult. These circulating ECFC are found at higher frequencies during neonatal life compared to adult life [46], with increased mobilization in the event of bony injury [47, 48].

In cocultures, ECFC organized into islets that with decreasing cluster sizes from Day 7 onward formed tubular structures that increased branching complexity over time during MSC coculture, while on their own, EPC monocultures were unable to form vascular networks on a collagen matrix. Establishment of more complex vessel-like networks was seen when ECFC/hfMSC were cocultured within 3D macroporous scaffolds in BM culture. This suggests that MSC have a critical support function on EC survival and sprouting in vitro, in agreement with findings from other groups [15, 49]. Subcutaneous implants of these coculture constructs in mice further supported the observations in vitro, with more extensive neovascularization of the scaffolds at 3 and 8 weeks postimplantation compared to MSC alone. Human-murine chimeric vessels were seen on confocal sections and species-specific immunohistochemistry only in ECFC/hfMSC cocultures, with human-specific CD31-positive structures arranged in a lumen-like fashion. This finding is consistent with previous reports [16, 17, 23, 45, 49].

By probing the transcriptome and secretome of UCB-ECFC, we identified several proinflammatory and angiogenic cytokines, in particularly Angiogenin, Angiopoietin-2, and PDGF-BB that were found in relatively high concentrations that could have contributed to this in vivo observation of increased angiogenesis. Inflammatory cytokines such as IL-1, IL-6, and TNF-α, and other angiogenic protagonists such as VEGF, bFGF, angiopoietins, PDGF-BB—angiogenin, IL-8, bFGF, FGF-4, TGF-β, and VEGF [50], have a direct stimulatory effect on angiogenesis in vivo. Such increase in vascularization could make a direct contribution to bone formation, as evident in the increased ectopic bone formed in the EPC/MSC constructs, suggesting an interdependent relationship between vascularization and bone formation. Usami et al. found a 1.6-fold higher capillary score formation on both the surface and central regions of canine-EPC/MSC scaffolds, accompanied by a 1.3-fold increase in bone area regenerated in subcutaneous implants [14], while Seebach et al. demonstrated increased early vascularization in the first 4 weeks with improved bone regeneration and healing in a rat critical-sized femoral defect model after EPC/MSC implantation compared to MSC alone [13].

CONCLUSIONS

In addition to their vasculogenic potential, circulating ECFC in late fetal/neonatal life have significant osteogenic-inducing capacity, which they effect through secretion of potent osteogenic regulators from the TGF-β superfamily. The interaction of UCB-ECFC with hfMSC allows both the generation of stable functional blood vessels and the direct potentiation of hfMSC osteogenic differentiation through secreted factors. These effects alone or in combination are likely to lend themselves toward applications in regenerative medicine and tissue engineering.

Acknowledgements

We acknowledge Tan Lay Geok and Dedy Sandekin for their kind assistance with the cell cultures and Nuryanti Binti Johana for the assistance with the qRT-PCR studies. This work was supported by National Medical Research Council of Singapore (NMRC/1179/2008 and NMRC/1268/2010). M.C. and J.K.Y.C. received salary support from NMRC Clinician Scientist Award (CSA/007/2009 and CSA/012/2009), respectively.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Ancillary